Synaptics® CX20921 Dual-Mic Far-Field Voice Processor SoC
AudioSmart® Product Brief
Publishe d November 7, 2017. Copyright © 2017 Synaptics Incorporated. All r ights reserved. PN: 190-000283-0 1 Rev. A
Standalone 2-Mic Far-Field Subsystem with
Integrated Low-Power Wake-on-Voice (WoV)
The Synaptics CX20921 is a high-performance, far-field voice
input processor System-on-Chip (SoC). The CX20921 is the
first product in this category with an integrated voice trigger
function that supports a low system power Wake-on-Voice
(WoV) function. The device includes two 24-bit microphone
Analog-to-Digital Converters (ADCs), a powerful dual core 32-
bit hardware Digital Signal Processor (DSP), and integrated
power management.
Combined with Synaptics AudioSmart® far-field voice input
processing software suite, including Smart Source Pickup
(SSP) noise suppression and full-duplex stereo Acoustic Echo
Cancellation (AEC), the CX20921 is the ideal turnkey solution
for adding voice control, voice search, and Voice over Internet
Protocol (VoIP)/speakerphone functionalities to SmartTV, Set-
Top Box (STB), Smart appliances, digital assistants, and
automotive hands-free control and communication systems.
The CX20921 device supports a microphone array with up to
two synchronized ADCs and programmable preamplifiers.
Each pre-amplifier is paired with a dedicated bias supply to
eliminate crosstalk. The entire input path guarantees 106dB
dynamic range, which maximizes speech-to-noise ratio with
low boost and prevents microphone saturation.
System Block Diagram
Synaptics’ proprietary far-field AudioSmart software leverages
the built-in hardware to deliver a superior voice interactive
experience in the most challenging room environments.
The voice processing chain also supports multiple modes of
processing, which are independently optimized for voice
communication and automatic speech recognition.
When ADC and DSP are active in the Wake-on-Voice (WoV)
mode, the device consumes less than 70mW, which is well
below the 0.5W EC 1275/2008 requirement for standby power.
Key Feat ures
• Low-power WoV mode with embedded third-party voice
wake-up engine
• Enables high-performance Automatic Speech Recognition
(ASR) and excellent voice clarity with only two mics using
Synaptics’ AudioSmart voice input processing software,
which includes:
o SSP noise suppression
o Full-duplex stereo AEC
• Compatible with multiple leading embedded and server-
based speech recognition vendors
• Dual far-field, 24-bit, 106dB dynamic range ADCs with mic
pre-amplifier:
o Enables low mic gain for far-field signals without
saturation for loud near field signals
Applications
• Voice-controlled Smart TV/STB
• Smart Bluetooth/WiFi speaker
• Voice interactive smart appliances
• Internet of Things (IoT) devices
• Automotive hands-free control/communication